Study On Ship Route Prediction In The Controlled Waterways Based On Trajectory Clustering

The traffic conditions in the controlled waterways are poor,and ships need to pass through the controlled waterways in sequence according to the command signals given by the traffic command system.An important basis for determining the order of ship traffic is the time when ships pass through the controlled waterways.At present,the channel centerline is selected in the command system to estimate the ship’s travel time.However,when the ship actually passes through the controlled waterways,it will choose the appropriate route according to its own load and flow conditions,and the difference between different routes is large.Therefore,there is a large error between the estimated time based on the channel centerline and the actual time,which seriously affects the accuracy of the system to reveal the pass signal.The key to the prediction of ship travel time is the accuracy of the ship’s route prediction in the controlled waterways.Therefore,improving the accuracy of ship route prediction in the controlled waterways is of great significance to vessel traffic scheduling.Taking the Shenbeizui controlled waterways in the upper reaches of the Yangtze River as research background,the thesis makes an in-depth analysis of the historical routes of ships in the region,explores the typical characteristics of ships during historical navigation,and studies how to improve the accuracy of ship route prediction in order to improve the accuracy of the ship’s transit time prediction and to provide a reliable basis for the ship’s traffic command system.The main research contents of the thesis is summarized as follows:(1)With regard to the problems of poor quality and inconsistent data format of ship trajectory data directly obtained from the control command system of the controlled waterways,preprocessing the missing values,outliers and data formats in ship trajectory data to obtain high-quality ship historical route data sets,and provide reliable data support for ship trajectories clustering.(2)Aiming at the shortcomings of the traditional Hausdorff distance,which can only measure the similarity between trajectory positions,an improved spatial trajectory similarity measurement method of Hausdorff distance is proposed,which not only considers the similarity between ship positions,but also takes the similarity between speeds into account.In view of the difficulty in determining the parameters in the DensityBased Spatial Clustering of Application with Noise(DBSCAN)based on density clustering,a gray wolf optimization algorithm is proposed to automatically optimize the parameters in the DBSCAN algorithm in order to achieve ship route clustering.Through simulation and experimental analysis,the effectiveness of the ship trajectories clustering method is verified.(3)Aiming at the problem that the current command system predicts the accuracy of ship travel time based on fixed routes,a ship route prediction method based on trajectories clustering is proposed.The method extracts the typical characteristic trajectories of the ship sailing in the controlled waterways by clustering the historical trajectories of the predicted ship,matches the current trajectory of the ship with the characteristic trajectories,and predicts the selected route when the ship passes the controlled waterways.The ship travel time is calculated based on the predicted route.Through simulation and experimental analysis,the effectiveness of the ship trajectories prediction method is verified.